The Equity Window Has Two Doors: Anthropic's $1.5 Trillion Test and a $91 Million Bank That Already Cleared

The headline statistic is a mismatch, not a contradiction. Just 121 IPOs priced through early November — 36% fewer than at the same point last year — and they raised $148.6 billion, 314% more than last year. The average deal this year ran about $1.2 billion, which is why the Renaissance IPO Index still fell 6.8% year to date while the S&P 500 rose 8.2%: the transactions that cleared got much bigger, and the ones an ordinary investor could actually buy got worse. What closed was the middle of the market, not the top.

Now the part the index cannot see. When Renaissance Capital flagged Narragansett Bancorp in its Nov. 6 weekly recap, it said the $91 million raise was not included in its IPO statistics — because a mutual holding company conversion is not an IPO by the index's definition. Add BCB Bancorp's $98.0 million underwritten offering and the arithmetic gets strange: a $91 million bank transaction cleared comfortably in a week when the index was down 6.8%. The equity window is not one market. It is a barbell, and the small end of it is being counted somewhere else.

Narragansett Bancorp raised real capital for the bank and sold it to the bank's own depositors. American Savings Bank raised nothing at all and existed to give its legacy shareholders a market price. BCB Bancorp raised capital into a balance sheet that had just lost $132.5 million. Mixing the three structures produces the wrong trade, because one is a capital return, one is a liquidity event for a selling shareholder, and one is a recapitalization after a credit blow-up.

The structure is the whole story. Narragansett Bancorp is the stock holding company for BayCoast Bank, a Massachusetts-chartered savings bank founded in 1851 with 25 branches across Massachusetts and Rhode Island and $2.96 billion of assets at Sept. 30. Before the deal its mutual holding company owned 100% of it, and as of Sept. 21 the company had no common stock outstanding at all. On Nov 2 the bank sold 9,098,800 shares at $10.00 — $91.0 million gross, including 761,760 shares for its employee stock ownership plan — leaving 21,160,000 shares outstanding: 55% with the mutual holding company, 43% with depositors and 2% with the charitable foundation. Piper Sandler was the agent, and the order book was oversubscribed by depositors eligible as of May 31, 2025.

What that structure produced on valuation is unusual. At June 30 the company held $189.2 million of stockholders' equity against $27.9 million of goodwill and other intangibles, so tangible common equity was about $161.3 million. An $211.6 million market cap on that base is 1.31 times tangible book — before the $91 million of proceeds even landed. Add the raise and pro-forma tangible book works out near $11.9 a share on 21.16 million shares, and the stock closed Nov 6 at $10.84 after opening at $10.80 on day one, trading at about 0.9 times pro-forma tangible book.

The balance sheet explains why this is capital recycling rather than a growth story. At June 30 equity was 6.6% of assets, and the raise lifts pro-forma equity to roughly 9% — while the loan book stays put. Net income was $4.42 million in the third quarter of 2026 against $3.5 million in the second, and interest expense fell from $14.3 million in the first quarter to $13.8 million in the second. Run the arithmetic and the dilution is explicit: annualized third-quarter earnings of about $17.7 million on roughly $280 million of pro-forma equity is a 6% return on equity, against about 9% before the raise.

No one in the market turned the key. The FDIC did. On Sept. 17 Chairman Travis Hill proposed a new merger framework with rapid, expedited and standard processing tracks, and made the smallest transactions eligible for final action in as little as five days. The agency's own numbers show the shift: an average of 107 days from receipt to final action in 2023 and 2024, 80 days in 2025, and 64 days year to date in 2026. The OCC had already restored expedited review in May 2025 and the FDIC had rescinded its 2024 policy statement on mergers two months later. Cutting the wait short makes sub-$3 billion bank deals worth doing, because a seller can no longer discount a 100-day regulatory queue into the price.

That is the mechanism behind the $91 million deal, and it runs well beyond Narragansett Bancorp. 118 FDIC-insured banks have been absorbed in twelve months, taking $436 billion of assets with them at a median target size of $376 million, while the median announcement-to-close time has fallen from 187 days in 2024 to 94 days so far this year. A $2.96 billion-asset thrift with $2.57 billion of deposits is no longer a company that has to be handed to a strategic buyer at a discount to compensate for the wait.

Piper Sandler, which underwrote the Narragansett offering, publishes the industry's pricing benchmark: the median price paid for a bank seller is about 141% of tangible book value. Bank Director's survey of bankers runs the other way — 68% of respondents would require at least 175% of tangible book to sell. The spread between what acquirers pay and what sellers insist on is where the profit is made, and Bain & Co. found that deals combining scale with new capabilities delivered total shareholder returns 14 to 18 percentage points above scale-only deals over the past two years.

For Narragansett Bancorp that spread is the option. A buyer taking the whole company at the market's 141% median would pay roughly $350 million — about 12% of assets, and a 5% earnings yield on a franchise currently earning around 0.6% on assets. That is the arithmetic that makes a $91 million offering worth doing at all: the depositors are funding a bank that has just printed a public price on a business somebody might want, and the acquirers holding 1.5 to 2.2 times tangible book can pay 141% and still improve their own returns on equity.

The tell is who got paid. Piper Sandler was the agent on the Narragansett offering under an agency agreement dated Aug. 11 and led American Savings Bank's IPO after the OCC declared the national charter effective on Sept. 15. It also publishes the 141%-of-tangible-book benchmark the market prices bank sellers against. Its own numbers show how narrow the equity franchise has become: third-quarter 2026 revenue of $443.7 million, down 4.1% from $462.4 million a year earlier, and diluted earnings of $0.78 a share against $0.845 — with the stock at $65.92, 31% below its $95.07 52-week high, on 15.5 times trailing earnings. Shows that bank equity issuance is the one part of the window that is growing while the generalist markets business shrinks.

The other upstream is the seller's own balance sheet. Hawaiian Electric Industries held 9.9% of American Savings Bank, a stake it could not simply unload in the open market; converting the institution to a national charter and listing it was the only way to manufacture a price for it. It sold into the IPO at $16.00 and left a publicly traded bank behind. HE's own shares have had a far worse year — $8.42 against a 52-week high of $17.38, 0.81 times book, and no dividend paid since September 2023. The bank IPO was a liquidity event for the holding company first and a growth story for the bank second.

Anthropic is expected to list on Nasdaq before Thanksgiving, with the roadshow starting the week of Nov. 9, and it is the only deal large enough to move the index. It is also being repriced downward in real time: investors were marking it near $2 trillion a week before the books opened and near $1.5 trillion after, with some buyers discussing $1.1 trillion, against a $965 billion private mark in May. The filing behind those numbers explains the pressure. Revenue grew 1,368% year over year in the quarter ended June 30, the operating loss for the first half was $693 million, adjusted operating income swung to $1.62 billion from a $1.85 billion loss a year earlier, and the headline first-half net loss widened to $80.7 billion on a non-operating item tied to convertible bonds. Anthropic lost $42 billion in 2025 and spent $7.33 billion on compute.

The point is not that AI issuance is impossible; it is that Anthropic is one deal, priced once, with no regulator obliged to make it easy. A five-day FDIC track, a 64-day review average and a Senate committee that can change hands in eight weeks make bank equity issuance a policy outcome rather than a market one. That is a different asset than a $1.5 trillion bookbuild, and it carries a different risk: the bank window depends on a policy with a shelf life.

The conclusion is narrow and testable. The equity window in November 2026 is not open or closed; it is a barbell with one enormous door and one very small one, and the small one is regulated. Makes a $91 million bank the only equity issuance a policy is actively creating, priced below tangible book while the firms that will eventually buy it trade at 1.5 to 2.2 times tangible book. The trade is not on AI issuance, which is priced once and marked down; it is on the pipeline of small banks that a five-day review track is turning from unsellable into sellable.